Security· August 17, 2026 at 12:51 a.m.
Chinese AI Company Zhipu Launches New Model Claiming Bug-Finding Superiority Over Anthropic, OpenAI
Key takeaways
- Zhipu introduces new AI model GLM-5.3
- GLM-5.3 outperforms Fable 5 and GPT-5.6 Sol on CyberGym benchmark
- 2,436 vulnerabilities discovered across 269 projects
Chinese tech company Zhipu has introduced a new AI model called GLM-5.3, which the firm claims outperforms American models in bug-finding capabilities. The announcement includes benchmark data showing that GLM-5.3 surpasses Fable 5 and GPT-5.6 Sol on the CyberGym benchmark, a test of a model's ability to solve real-world cybersecurity challenges. Zhipu claims that the model discovered 2,436 vulnerabilities across 269 projects, including 1,097 medium-to-high severity issues, in tests conducted with Chinese companies on real-world codebases. The findings span various areas such as system kernels, operating systems, browser engines, open-source infrastructure, web applications, and network protocols. GLM-5.3 performed worse than western models on other security and coding benchmarks, but its bug-finding prowess signals China's growing capability in cybersecurity.
